Early genome-wide association studies (GWAS) focused on the association between one or a small number of single-nucleotide polymorphisms (SNPs) and a phenotype. Such studies were likely to miss associations where a large number of SNPs have a mild influence on the phenotype. Researchers have begun to look for associations based on sets of SNPs corresponding to pre-defined or learned gene sets. Finding a robust aggregation function that that quantifies the relationship between a set of SNPs and a phenotype has been elusive.
The subject matter claimed herein is not limited to embodiments that solve any disadvantages or that operate only in environments such as those described above. Rather, this background is only provided to illustrate one exemplary technology area where some embodiments described herein may be practiced.